Induction of aquaporin 4-reactive antibodies in Lewis rats immunized with aquaporin 4 mimotopes

Abstract

Most cases of neuromyelitis optica spectrum disorders (NMOSD) harbor pathogenic autoantibodies against the water channel aquaporin 4 (AQP4). Binding of these antibodies to AQP4 on astrocytes initiates damage to these cells, which culminates in the formation of large tissue destructive lesions in the central nervous system (CNS). Consequently, untreated patients may become permanently blind or paralyzed. Studies on the induction and breakage of tolerance to AQP4 could be of great benefit for NMOSD patients. So far, however, all attempts to create suitable animal models by active sensitization have failed. We addressed this challenge and identified peptides, which mimic the conformational AQP4 epitopes recognized by pathogenic antibodies of NMOSD patients. Here we show that these mimotopes can induce the production of AQP4-reactive antibodies in Lewis rats. Hence, our results provide a conceptual framework for the formation of such antibodies in NMOSD patients, and aid to improve immunization strategies for the creation of animal models suitable for tolerance studies in this devastating disease.

Keywords: Animal model; Antibodies; Aquaporin 4; Infections; Mimotopes; Neuromyelitis optica spectrum disorders.

Fig. 1

Characterization of NMO-IgGs used for mimotope search. a-d Immunofluorescence staining of Lewis rat astrocytes and analysis by confocal microscopy. The NMO-IgG preparation IV containing pathogenic AQP4-reactive antibodies recognizing conformational epitopes on the surface of astrocytes (a, red), a commercial AQP4-reactive antibody recognizing intracellular AQP4 epitopes (b, green) and an antibody directed against GFAP (c, blue) were used for stainings. Stainings against surface and intracellular AQP4 epitopes were merged to prove that IV contains AQP4-reactive antibodies reacting with rat AQP4 (d, white). e-f Formation of astrocyte-destructive lesions in experimental NMO. Shown here are spinal cords of Lewis rats injected with myelin basic protein-specific T cells and the NMO-IgG preparations IV (e), III (f) and I (g). Sections were stained with antibodies against AQP4 to show astrocytes (brown) and counterstained with hematoxylin to show nuclei (blue). h For each NMO-IgG preparation used, the phage display peptide library Ph.D.-12 was subjected to three rounds of negative selection on human control-IgG (Subcuvia) to deplete phages binding to “common antibodies”, and of positive selection on NMO-IgG to enrich for phages binding to the AQP4-reactive antibodies contained within the NMO-IgG preparation. At the end of these selections, bound phages were released, amplified, and sequenced for the identification of the mimotopes. i Example of sequencing results for mimotope IV-04. The DNA sequence represents the genomic (+) ssDNA in 5´➔ 3´ direction. Underneath you see the corresponding amino acid sequence (capital letters). Mimotope flanking regions are shown in gray, restriction enzyme recognitions sites in yellow and red, and the mimotope sequence in magenta

Fig. 2

Characterization of phage clones, phage-displayed peptides, and mimotopes. a-c For different NMO-IgG preparations, different experimental approaches were made to narrow down the number of phage-displayed peptides for further studies. a ELISA with single phage clones to evaluate interactions with the NMO-IgG preparation I. Single phage clones after 3 rounds of negative/positive selection (I-01 - > I-19) and 5 randomly picked phage clones without preceeding selection (L7 - > L11) were tested for their ability to react with the NMO-IgG preparation I, BSA, or control-IgG (Subcuvia). Each sample contained 10⁸ phages. Data represent three different experiments and are shown as mean + SEM (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, detected with one-way ANOVA followed by Sidak’s multiple comparisons test). b ELISA to verify that peptides, but not phage particles bind to the NMO-IgG preparation IV. IV was pre-incubated with specific, random, or no synthesized peptides prior to the incubation with peptide-displaying phages. Bound phages were then detected with horse radish peroxidase-conjugated anti-M13 antibodies and TMB substrate, and the absorbance was measured at 450 nm in an ELISA reader. Data represent triplicates of one experiment and are shown as mean + SEM. c Phage-displayed peptides identified with the NMO-IgG preparation III were mapped onto the 3D structure of human AQP4 (protein data base (pdb) accession number 3GD8) using PepSurf. Extracellular loops are displayed on top of the structure, intracellular parts of the molecule on the bottom of the 3D structure. The final alignment represents the best possible path in a defined surface graph. The probability obtaining the same alignment with a random sequence is given by the corresponding p-value. III-01 (P-value: 0.00017), III-04 (P-value: 0.00106), III-06 (P-value: 0.00007), and III-09 (P-value: 0.00028) mapped at least partially to the extracellular loops of AQP4 (red), while III-10 (P-value: 0.00008) and III-17 (P-value: 0.00047) mapped to intracellular or helical structures, respectively. d Selected peptides were tested for their ability to interfere with the binding of AQP4-reactive antibodies of NMO-IgG preparations to AQP4. Flow cytometry of AQP4 M23-transfected HEK293A cells reacting with NMO-IgG preparations V, VI, and III pre-incubated with the indicated peptides. Pre-incubation without peptide, or with a random peptide (SPRAISSYPLNEGGGS) served as negative controls. The data shown here are the mean values (+/−SEM) of the percentage of NMO-IgG binding after pre-incubation with peptides, obtained from 4 (V and III) or 5 (VI) independently performed experiments. Please note that the binding of the NMO-IgG preparations pre-incubated without peptide or with random peptides slightly differed from each other. Therefore, we referred to the binding of NMO-IgG without peptide as 100% (dashed red line) and always also show the percentage of binding of NMO-IgG pre-incubated with random peptide (solid red line). The lower one of these two different values was used as reference for the percentage of blocking achieved with the different peptides. Statistics was calculated using one-tailed, Welch-corrected t-tests. Blue arrows indicate mimotopes used for immunization

Fig. 3

Detection of AQP4-specific antibodies in sera of Lewis rats immunized with mimotopes. Life-cell immunofluorescence staining against the surface antigens rat AQP4-EmGFP and human AQP4-EmGFP was made. Antigens were transiently transfected into HEK293 cells (green) and sera of Lewis rats immunized with mimotopes IV-04, III-01, I-13, I-04, or IV-38 were added. Bound antibodies were then detected with anti-rat-IgG (red). Merge of green and red signals reveals surface staining of transfected cells in yellow if antibody signals were strong. Additional images at 40x magnification were added for details (see inlay). For titer values see Table 3

Fig. 4

Antibody mediated complement activation of serum samples from rats immunized with AQP4 mimotopes. a: Serum AQP4 antibodies from an AQP4 antibody positive NMOSD patient (a) and the mouse monoclonal E5415A AQP4 antibody were able to activate the complement cascade in the presence of active rat complement on rat-AQP4-EmGFP (green) expressing HEK293 cells resulting in TCC deposition (red) and cell death (DAPI, blue). a and b are positive controls (pos co 1 and 2). c-f: Sera from rats immunized with mimotope-AQP4_268–285/CFA (mimotopes were IV-04 inducing rat AQP4-spec antibodies in a titer of 1:160, III-01 inducing rat AQP4-spec antibodies in a titer of 1:160, I-04 inducing rat AQP4-spec antibodies in a titer of 1:320, and I-13 inducing rat AQP4-spec antibodies in a titer of 1:320) were not able to activate the complement cascade as shown by fewer DAPI (blue) positive cells and no TCC (red) deposition. g: Serum from a III-01-AQP4_268–285/CFA immunized rat with an antibody titer of zero was used as a first negative control (neg co 1) and shows no antibody mediated complement activation. h: Same serum as in A in the presence of heat-inactivated rat complement (second negative control (neg co 2) does not activate the complement cascade and shows no TCC (red) formation. Scale bar = 100 μm